Hi Lars - I see what the problem is.

I actually gave you some bad advice, I forgot that you had a moebius topology here. Sorry about that!

A periodic surface has the kind of smooth closure that you need (the surface is defined with a set of points that overlap with one another in the closure area) but a periodic surface is usually set up in NURBS such that the surface is joined with the same sides to the "outside" where it is touching itself.

Like for example if this is the UV layout of a typical periodic surface, the 2 points circled in red would be touching one another in 3D along the seam.



But for a moebius topology I guess you need a kind of special case periodic surface that would match things opposite from that, like match one corner to the one across the diagonal from it instead of straight across. I don't know of any existing command that would do that, it would take some specialized code.

I don't think that it is going to be very easy to get a proper smooth single surface out of this thing since NURBS is not generally set up with this kind of topology in mind.

Probably your best bet to produce the final form would be to cut back the area around that seam a bit to make an opening, then use Blend to put in a smooth piece. That should have a lot better chance at giving a thickened result I think.

The way you would do that is in Rhino use ShowEdges to find the seam edge, then use Split with the Isocurve option to place the cut lines near there, to create the opening. Then I took it into MoI, did a Blend with Continuity:Curvature option, then joined the pieces, then Shell, and then there is a problem with the result where there is one of those internal walls created since Shell is not expecting this kind of thing, so you have to use Edit/Separate, delete the internal walls, then rejoin and finally there is what seems to be a proper solid result which I've attached here (doublemoebius_blend_3dm.zip)

Hope this helps!

- Michael